Skip to content

Dataset · timbrooks

Image pairs with editing instructions

timbrooks/instructpix2pix-clip-filtered

CLIP-filtered dataset of 100k to 1M examples for training text-based photo editing.

What's inside

Multimodal dataset of images and English text containing between 100,000 and 1,000,000 records. Each row includes an original image (original_image), an editing instruction (edit_prompt), and the resulting edited image (edited_image), filtered using CLIP.

What you can do with it

  • Training diffusion models and LoRA to edit images using text prompts
  • Testing multimodal models on their understanding of image editing instructions
Languages
English
Size
100K–1M rows
Downloads this month
5.1K
Likes
48

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("timbrooks/instructpix2pix-clip-filtered")